Trajectory Planning for Multi-UAV Cooperative Pursuit-Evasion Game Based on Sequential Convex Programming

Current solution methods for multi-pursuer-one-evader differential games, such as Voronoi partition-based approaches and differential game theory, typically require solving complex partial differential equations with high computational complexity. Efficient solutions are particularly lacking for fixed-wing UAVs with holonomic kinematic constraints. This paper proposes a differential game method based on Sequential Convex Programming (SCP). The proposed method decouples the coupled pursuit-evasion differential game into several differential game subproblems and transforms the non-convex evasion objective function into anti-capture constraints. By linearization, these constraints are convexified and then converted into convex constraint penalties using the penalty function method. Consequently, the differential game subproblems are transformed into multiple sequential convex programming problems, enabling efficient and rapid computation of approximate saddle-point solutions. Simulation experiments conducted in multi-pursuer-one-evader game scenarios demonstrate the effectiveness of the proposed method.